Immune gene expression in head and neck squamous cell carcinoma patients

Immune gene expression in head and neck squamous cell carcinoma patients

European Journal of Cancer 121 (2019) 210e223 Available online at www.sciencedirect.com ScienceDirect journal homepage: www.ejcancer.com Original R...

1020KB Sizes 6 Downloads 24 Views

European Journal of Cancer 121 (2019) 210e223

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.ejcancer.com

Original Research

Immune gene expression in head and neck squamous cell carcinoma patients Charlotte Lecerf a,*,1, Maud Kamal a,**,1, Sophie Vacher b, Walid Chemlali b, Anne Schnitzler b, Claire Morel a, Coraline Dubot a, Emmanuelle Jeannot b,c, Didier Meseure c, Jerzy Klijanienko c, Odette Mariani c, Edith Borcoman d, Valentin Calugaru e, Nathalie Badois f, Anne Chilles e, Maria Lesnik f, Samar Krhili f, Olivier Choussy f, Caroline Hoffmann d,f, Eliane Piaggio d, Ivan Bieche b,g, Christophe Le Tourneau a,h,i a

Department of Drug Development and Innovation, Institut Curie, Paris & Saint-Cloud, France Department of Genetics, Institut Curie, PSL Research University, Paris, France c Department of Pathology, Institut Curie, PSL Research University, Paris, France d INSERM U932 Research Unit, Institut Curie, PSL Research University, Paris, France e Department of Radiotherapy, Institut Curie, PSL Research University, Paris, France f Department of Surgery, Institut Curie, PSL Research University, Paris, France g EA7331, Paris Descartes University, Faculty of Pharmaceutical and Biological Sciences, Paris, France h INSERM U900 Research Unit, Institut Curie, Saint-Cloud, France i Versailles-Saint-Quentin-en-Yvelines University, Montigny-le-Bretonneux, France b

Received 6 February 2019; received in revised form 3 July 2019; accepted 30 August 2019 Available online 5 October 2019

KEYWORDS Head and neck squamous cell

Abstract Background: Nivolumab and pembrolizumab targeting programmed cell death protein 1 (PD-1) have recently been approved among patients with recurrent and/or metastatic head and neck squamous cell carcinoma (HNSCC) who failed platinum therapy. We aimed to

* Corresponding author: ** Corresponding author: E-mail addresses: [email protected] (C. Lecerf), [email protected] (M. Kamal), [email protected] (S. Vacher), walid.chemlali@ curie.fr (W. Chemlali), [email protected] (A. Schnitzler), [email protected] (C. Morel), [email protected] (C. Dubot), emmanuelle. [email protected] (E. Jeannot), [email protected] (D. Meseure), [email protected] (J. Klijanienko), [email protected] (O. Mariani), [email protected] (E. Borcoman), [email protected] (V. Calugaru), [email protected] (N. Badois), anne. [email protected] (A. Chilles), [email protected] (M. Lesnik), [email protected] (S. Krhili), [email protected] (O. Choussy), [email protected] (C. Hoffmann), [email protected] (E. Piaggio), [email protected] (I. Bieche), christophe.letourneau@curie. fr (C. Le Tourneau). 1 Contributed equally and should be considered co-first authors. https://doi.org/10.1016/j.ejca.2019.08.028 0959-8049/ª 2019 Elsevier Ltd. All rights reserved.

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

carcinoma; Immune checkpoints; Prognostic biomarker; Gene expression

211

evaluate the prognostic value of selected immune gene expression in HNSCC. Patients and methods: We retrospectively assessed the expression of 46 immune-related genes and immune-cell subpopulation genes including immune checkpoints by real-time polymerase chain reaction among 96 patients with HNSCC who underwent primary surgery at Institut Curie between 1990 and 2006. Univariate and multivariate analyses were performed to assess the prognostic value of dysregulated genes. Results: The Median age of the population was 56 years [range: 35e78]. Primary tumour location was oral cavity (45%), oropharynx (21%), larynx (18%) and hypopharynx (17%). Twelve patients (13%) had an oropharyngeal human papillomavirusepositive tumour. Most significantly overexpressed immune-related genes were TNFRSF9/4-1BB (77%), IDO1 (75%), TNFSF4/OX40L (74%) and TNFRSF18/GITR (74%), and immune-cell subpopulation gene was FOXP3 (62%). Eighty-five percent of tumours analysed overexpressed actionable immunity genes, including PD-1/PD-L1, TIGIT, OX40/OX40L and/or CTLA4. Among the immunerelated genes, high OX40L mRNA level (p Z 0.0009) and low PD-1 mRNA level (p Z 0.004) were associated with the highest risk of recurrence. Among the immune-cell subpopulation genes, patients with high PDGFRB mRNA level (p < 0.0001) and low CD3E (p Z 0.0009) or CD8A mRNA levels (p Z 0.004) were also at the highest risk of recurrence. Conclusions: OX40L and PDGFRB overexpression was associated with poor outcomes, whereas PD-1 overexpression was associated with good prognosis in patients with HNSCC treated with primary surgery, suggesting their relevance as potential prognostic biomarkers and major therapeutic targets. ª 2019 Elsevier Ltd. All rights reserved.

1. Introduction Head and neck squamous cell carcinoma (HNSCC) is the seventh cause of cancer with a yearly 40e50% mortality [1]. Classical risk factors for HNSCC include tobacco and alcohol consumption, as well as human papillomavirus (HPV) infection that has been demonstrated to have a prognostic impact [2]. The Cancer Genome Atlas reported the genomic landscape of more than 270 primary HNSCCs [3] with mutations in several oncogenes including PIK3CA (21%) and HRAS (4%), as well as in tumour suppressor genes including TP53 (72%), CDKN2A (22%), FBXW7 (5%), KMT2D (MLL2) (18%) and PTEN (2%) [3,4]. Genomic alterations involving the cell cycle (TP53, CCND1, CDKN2A), as well as FGFR1 amplifications, and tumour genomic alterations burden were shown to be prognostic and potential therapeutic targets for patients with HNSCC [5]. No relevant biomarkers for tailored therapeutic strategies have been identified in HNSCC to date. Beside surgery, radiotherapy and chemotherapy, HNSCC treatment includes targeted therapy and immunotherapy. Cetuximab, a monoclonal antibody that targets epidermal growth factor receptor (EGFR), has been the first targeted therapy approved in HNSCC, both in the locally advanced setting combined with radiotherapy [6] and in the first-line recurrent and/or metastatic setting in combination with chemotherapy [7]. Two antieprogrammed cell death protein 1 (PD-1) immune checkpoint inhibitors have been approved for the treatment of recurrent and/or metastatic HNSCC

refractory to platinum therapy in 2016 [8,9]. These agents are better tolerated than chemotherapy and demonstrated durable responses in a minority of patients [8,9]. At the tumour microenvironment (TME) level, the infiltration of HNSCC by innate and adaptive immune cells has been well documented. Several studies have identified immune cells with a prognostic value, such as CD8þ T cells, Foxp3þ regulatory T cells. The presence of tertiary lymphoid structures was also reported to affect prognosis [10e14]. OX40, PD-1 and CTLA4 were shown to have a significantly higher expression in T-cell subsets isolated from tumours of patients with HNSCC [15]. Few integrative studies reported the prognostic value of immune genes in HNSCC. We aimed in this study to assess the expression of immune genes and to evaluate their prognostic value in patients with HNSCC who are untreated.

2. Patients and methods 2.1. Patients We retrieved samples from patients with HNSCC who underwent upfront surgery at the Institut Curie between 1990 and 2006. We selected 96 patients with complete clinical, histological and biological data and long-term follow-up. This study was approved by the internal review board of Institut Curie and was conducted in accordance with the ethics principles of the Declaration of Helsinki. In

212

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

accordance with the French regulations, all patients were informed that analyses were to be performed on the biological specimens obtained during their treatment, and they did not express their opposition. 2.2. Gene selection Forty-six genes involved in the immune process were selected, including 30 genes defined as immune-related genes and 16 genes that were defined as immune-cell subpopulation genes (Supplementay Table 1). We chose TBP (Genbank accession number NM_003194) which encodes the TATA boxebinding protein as an RNA control gene. 2.3. DNA sequencing and mutation assessment Targeted DNA sequencing of a selection of 100 genes corresponding to the most frequently altered genes in HNSCC and potential therapeutic targets was performed on Illumina HiSeq2500 sequencer and then annotated in the COSMIC and 1000 genome databases [16], as described in the study by Dubot et al [5]. Tumour genomic alteration burden was assessed by summing the number of recurrent deleterious genomic alterations (single nucleotide nariation (SNV) þ copy number variantion (CNV)) per sample. Sanger sequencing was also performed to confirm PIK3CA, KRAS, HRAS and NRAS mutations. 2.4. Human papillomavirus genotyping HPV status was assessed at the Pathology Department of the Institut Curie. HPV typing was conducted using total DNA isolated from formalin-fixed tissue blocks. Real-time polymerase chain reaction (RT-PCR) was performed with Sybr Green and specific primers for HPV16 and HPV18 using a 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster City, CA). 2.5. Real-time quantitative polymerase chain reaction PCR consumables, RNA extraction, ctDNA synthesis and PCR reaction conditions were previously described in detail [17]. Primers are described in Supplementary Table 2. For each investigated gene, mRNA values  3 were considered as overexpression and 0.33 as underexpression. We previously used the same cut-off value for altered tumour gene expression [17].

Table 1 Clinical, biological and pathological characteristics of the 96 patients with HNSCC in relation with disease-free interval (DFI). Characteristics

Patients (%)

Total 96 (100) Age <56 46 (48) 56 50 (52) Gender Female 19 (20) Male 77 (80) Alcoholc Yes 50 (70) No 21 (30) Tobacco consumptiond Yes 58 (73) No 22 (27) HPV status Negative 84 (87) Positive 12 (13) UICC stage Stage I 10 (10) Stage II 15 (16) Stage III 12 (13) Stage IV 59 (62) Tumour location Oral cavity 43 (45) Oropharynx 20 (21) Larynx 17 (18) Hypopharynx 16 (17) Oncogenes’ mutational statuse Not mutated 78 (81) At least one mutated 18 (19) Number of molecular alterationsf <3 56 (60) 3 37 (40)

Eventsa (%)

DFIb

45 (47) 20 (44) 25 (50)

0.89 (NS)

8 (42) 37 (48)

0.71 (NS)

24 (48) 8 (38)

0.17 (NS)

28 (48) 7 (32)

0.075 (NS)

42 (50) 3 (25)

0.032

5 (50) 6 (40) 4 (33) 30 (51)

0.68 (NS)

22 (51) 5 (25) 8 (47) 10 (63)

0.053 (NS)

36 (46) 9 (50)

0.079 (NS)

27 (48) 17 (46)

0.082 (NS)

DFI: disease-free interval; NS: not significant; HNSCC: head and neck squamous cell carcinoma; UICC: Union for International Cancer Control; HPV: human papillomavirus. a Events: locoregional and/or metastatic recurrence, second cancer. b Log-rank test. c Alcohol use was considered at 10 gr/day or more (i.e. alcohol unit). Information was available for 71 patients. d Tobacco use was considered at 10 pack-years or more. Information was available for 80 patients. e PIK3CA, NRAS, HRAS or KRAS oncogenes. f Number of molecular alterations among a selection of 100 genes as previously determined in the study by Dubot et al. [5]. Information was available for 93 patients.

and methods). We performed immunohistochemical study of the tumour microenvironment by OX40L immunostaining in tumour cells, stromal cancer-associated fibroblasts (CAFs) and mononuclear inflammatory cells (MICs). All quantifications were performed with blinding of 2 expert pathologists to patient status.

2.6. Immunohistochemistry/OX40L protein expression 2.7. Statistical analyses We performed immunohistochemistry (IHC) assay by using the OX40L (rabbit, 59036, 1/100, pH Z 9; Cell Signaling Technology) antibody in a series of 20 HNSCCs among the 96 HNSCC patients, corresponding to 10 patients with high OX40L mRNA level and 10 with low OX40L mRNA level (Supplementary material

The clinicopathological features were tested for association with disease-free interval (DFI) by using the log-rank test. DFI was determined from the time of initial diagnosis to the time of the first event among locoregional recurrence, metastatic recurrence or second cancer. The

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

Fig. 1. Gene expression versus molecular alterations and clinical patient characteristics. HPV, human papillomavirus.

213

214

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

Table 2 mRNA expression of 30 immune-related genes relative to normal tissue mRNA level. Gene

Protein

HLA-DRA LAG3 PVR PVRIG TIGIT CD96 CD226 TNFSF9 TNFRSF9 TNFSF18 TNFRSF18 ICOSLG ICOS OX40L

Ligand Receptor Ligand Receptor Receptor Receptor Receptor Ligand Receptor Ligand Receptor Ligand Receptor Ligand

OX40 CD70 CD27 TIM-3 LGALS9 PD-L1 PD-L2 PD-1 ENTPD1 IDO1 NT5E TNFRSF14 CD276 CD80 CD86 CD28

Receptor Ligand Receptor Ligand Receptor Ligand Ligand Receptor Ligand Ligand Ligand Receptor Receptor Ligand Ligand Receptor

Alias

CD155 CD112R

CD137L, 4-1BBL CD137, 4-1BB GITRL GITR ICOSL, B7H2 TNFSF4, CD134L, CD252 TNFRSF4, CD134 TNFSF7 TNFRSF7 HAVCR2 GALECTINE-9 CD274 PDCD1LG2 PDCD1, CD279 CD39 CD73 HVEM B7H3 B7-1 B7-2

Head and neck normal tissuea

Head and neck squamous cell carcinomas

p-valueb

% under expression

% normal expression

% over expression

n Z 27

n Z 96

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

(0.01e5.87) (0.29e4.96) (0.66e3.93) (0.23e7.45) (0.04e9.66) (0.14e4.74) (0.49e4.16) (0.17e5.4) (0.23e8.39) (0e36.1) (0.02e4.21) (0.46e2.52) (0.11e6.99) (0e9.94)

0.74 2.01 1.22 1.45 2.77 1.12 0.85 1.54 5.05 1.81 4.13 0.64 3.45 6.52

(0.01e9.95) (0.2e16.7) (0.47e4.78) (0.17e8.13) (0.09e23.4) (0.12e7.28) (0.08e4.3) (0.14e10.1) (0.61e29.9) (0.05e58.4) (0.57e17.4) (0.1e7.98) (0.27e24.6) (0.24e48.0)

0.91 (NS) 0.001 0.28 (NS) 0.24 (NS) <0.0001 0.59 (NS) 0.05 (NS) 0.011 <0.0001 0.063 (NS) <0.0001 0.0003 <0.0001 <0.0001

41% 6% 0% 5% 2% 8% 10% 4% 0% 9% 0% 8% 1% 1%

47% 58% 95% 77% 51% 79% 88% 78% 23% 55% 26% 89% 44% 25%

13% 35% 5% 18% 47% 13% 2% 18% 77% 35% 74% 3% 55% 74%

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

(0.29e2.73) (0.25e4.32) (0.05e18.6) (0.35e2.03) (0.34e3.02) (0.31e2.79) (0.29e2.75) (0.22e8.09) (0.42e2.48) (0.09e8.68) (0.32e2.35) (0.42e1.6) (0.34e1.79) (0.16e5.5) (0.37e2.61) (0.31e6.49)

2.12 4.21 1.92 1.57 1.21 1.26 1.45 1.97 1.02 6.41 1.03 0.61 2.05 5.35 1.81 1.32

(0.17e13.0) (0.22e92.0) (0.04e14.3) (0.24e8.3) (0.1e10.2) (0.16e38.7) (0.27e13.6) (0.11e14.2) (0.2e3.26) (0.14e184.4) (0.06e6.35) (0.07e1.75) (0.26e7.61) (0.67e23.0) (0.27e5.98) (0.1e8.16)

<0.0001 <0.0001 0.11 (NS) 0.0006 0.31 (NS) 0.056 (NS) 0.035 0.019 0.93 (NS) <0.0001 0.79 (NS) 0.0006 <0.0001 <0.0001 <0.0001 0.36 (NS)

1% 1% 7% 1% 5% 6% 4% 6% 2% 2% 6% 18% 1% 0% 1% 5%

74% 34% 58% 78% 84% 70% 77% 62% 97% 23% 82% 82% 77% 20% 78% 79%

25% 65% 34% 21% 10% 24% 19% 32% 1% 75% 12% 0% 22% 80% 21% 16%

a Median (range) of gene mRNA levels; the mRNA values of the samples were normalised such that the median of the 27 head and neck normal tissues mRNA values was equal to 1. b Kruskal-Wallis H test.

clinicopathological and biological characteristics were tested for association with transcript-level expression by using chi-square tests for categorical variables. The association between clinical variables and RNA levels was tested using Kruskal-Wallis H tests. Cox proportional hazard regression was used to estimate hazard ratio (HR) and their 95% confidence intervals (95% CI) for covariates associated with DFI, showing significance at p < 0.1 on univariate analysis. Differences between two populations were judged significant at confidence levels greater than 95% (p < 0.05) [18]. Unsupervised hierarchical cluster analyses were performed using Morpheus algorithm to identify homogenous genes and tumour groups regarding molecular data.

3. Results

years [range: 35e78]. Most patients were male with tobacco and alcohol consumption. Twelve patients (13%) were HPV-positive, with a majority with oropharyngeal cancer. Pathological staging showed a high proportion of stage IV. The main tumour location was the oral cavity (45%), followed by the oropharynx (21%), larynx (18%) and hypopharynx (17%). Most patients had less than three tumour genomic alterations as previously determined [5]. HPV infection was the only characteristic that significantly impacted DFI (p Z 0.032, logrank test), with a higher DFI reported for HPV-positive patients. 3.2. mRNA expression of immune genes Fig. 1 illustrates mRNA expression of the 46 immune genes according to clinical and molecular characteristics of the 96 patients with HNSCC.

3.1. Patient characteristics The characteristics of the 96 patients with untreated HNSCC are listed in Table 1. The median age was 56

3.2.1. mRNA expression of the immune-related genes Among the 30 immune-related genes analysed, 18 genes were significantly deregulated in HNSCC tumours as

Table 3 mRNA expression of 30 immune-related genes in HNSCC according to HPV status and the number of molecular alterations. Gene

Ligand Receptor Ligand Receptor Receptor Receptor Receptor Ligand Receptor Ligand Receptor Ligand Receptor Ligand Receptor Ligand Receptor Ligand Receptor Ligand Ligand Receptor Ligand Ligand Ligand Receptor Receptor Ligand Ligand Receptor

Alias

HPV- versus HPVþ HPVa

HPVþ

n Z 84

n Z 12

0.75 (0.01e9.95) 1.88 (0.2e16.7) CD155 1.27 (0.53e4.78) CD112R 1.6 (0.17e8.13) 2.71 (0.09e23.4) 1.06 (0.12e7.28) 0.83 (0.08e4.30) CD137L 1.54 (0.14e10.1) CD137 5.05 (0.61e25.2) GITRL 2.030.05e58.4) GITR 4.26 (0.57e17.4) ICOSL 0.61 (0.10e2.01) 3.13 (0.27e24.6) TNFSF4 6.52 (0.24e48.0) TNFRSF4 2.12 (0.17e13.0) TNFSF7 4.10 (0.22e88.2) TNFRSF7 1.9 (0.04e14.3) HAVCR2 1.56 (0.24e8.30) GALECTINE-9 1.19 (0.10e10.20) CD274 1.17 (0.16e38.7) PDCD1LG2 1.42 (0.27e13.6) PDCD1 1.92 (0.11e14.2) CD39 1.02 (0.20e3.26) 6.41 (0.14e184.4) CD73 1.15 (0.18e6.35) HVEM 0.61 (0.07e1.75) B7H3 2.15 (0.26e7.61) B7-1 5.47 (0.67e23.0) B7-2 1.76 (0.27e5.98) 1.23 (0.10e8.16)

0.15 3.04 1.02 2.47 4.07 1.79 1.03 1.35 6.08 1.33 2.99 0.82 4.45 9.35 2.09 5.85 2.79 1.87 1.86 1.77 1.51 2.60 0.76 6.21 0.44 0.66 1.25 4.68 2.05 1.72

(0.02e6.14) (0.71e9.02) (0.47e3.51) (0.47e6.39) (1.09e10.7) (0.30e3.85) (0.32e2.61) (0.69e3.78) (2.32e29.9) (0.33e6.46) (0.89e9.66) (0.33e7.98) (2.05e7.77) (0.43e18.1) (0.94e4.67) (0.87e92.0) (0.49e5.75) (0.71e3.49) (0.26e5.38) (0.47e8.83) (0.29e6.27) (0.74e6.34) (0.41e1.83) (1.70e128.3) (0.06e2.87) (0.21e1.27) (0.59e4.81) (1.99e9.38) (0.77e3.09) (0.59e2.92)

p-valueb

0.69 (NS) 0.23 (NS) 0.41 (NS) 0.047 0.23 (NS) 0.27 (NS) 0.43 (NS) 0.71 (NS) 0.26 (NS) 0.36 (NS) 0.0097 0.10 (NS) 0.27 (NS) 0.91 (NS) 0.54 (NS) 0.34 (NS) 0.45 (NS) 0.89 (NS) 0.15 (NS) 0.21 (NS) 0.94 (NS) 0.23 (NS) 0.091 (NS) 0.84 (NS) 0.008 0.68 (NS) 0.005 0.59 (NS) 0.98 (NS) 0.47 (NS)

Not mutated versus mutated oncogenes

Number of molecular alterations <3 versus 3

Not mutated

Mutated

<3

3

n Z 78

n Z 18

n Z 56

n Z 37

0.73 2.03 1.20 1.31 2.59 1.04 0.76 1.60 4.83 1.85 4.23 0.57 3.40 5.85 2.12 4.41 1.82 1.52 1.22 1.17 1.41 1.92 1.02 6.27 1.15 0.59 2.07 5.23 1.71 1.17

1.11 1.78 1.25 1.80 4.01 1.28 0.88 1.40 7.21 1.49 3.74 0.88 4.03 7.39 2.27 3.93 2.64 2.02 1.19 1.72 1.84 2.18 1.03 7.83 0.89 0.70 1.91 7.06 2.26 1.54

0.78 2.79 1.32 1.93 3.95 1.64 1.24 1.56 6.43 1.73 3.70 0.73 4.49 7.94 2.23 4.31 2.79 2.01 1.38 1.72 1.79 2.67 1.17 8.35 1.16 0.71 1.99 6.20 2.29 1.85

0.65 1.20 1.07 0.91 1.83 0.79 0.49 1.52 4.39 2.44 6.80 0.54 2.13 5.45 1.70 4.07 0.93 1.16 0.89 0.74 1.17 0.97 0.76 4.45 0.99 0.45 2.13 3.66 1.36 0.90

(0.01e6.63) (0.20e15.6) (0.51e4.78) (0.17e8.13) (0.09e23.4) (0.12e7.09) (0.08e3.95) (0.14e6.27) (0.61e29.9) (0.16e58.4) (0.57e17.4) (0.10e7.98) (0.27e24.6) (0.24e39.5) (0.17e13.0) (0.22e88.2) (0.04e14.3) (0.24e5.63) (0.10e10.2) (0.16e10.9) (0.27e13.6) (0.11e12.2) (0.20e3.26) (0.14e138.9) (0.18e6.35) (0.07e1.53) (0.26e7.61) (0.67e23.0) (0.27e5.98) (0.10e8.16)

(0.03e9.95) (0.77e16.7) (0.47e4.33) (0.49e6.27) (0.67e13.1) (0.35e7.28) (0.37e4.30) (0.40e10.1) (1.33e19.3) (0.05e6.34) (0.72e11.0) (0.26e7.15) (0.37e9.28) (1.58e48.0) (0.94e4.40) (0.66e91.4) (0.13e5.75) (0.89e8.30) (0.59e5.04) (0.47e38.7) (0.29e10.2) (0.85e14.2) (0.45e1.93) (1.34e184.4) (0.06e4.95) (0.31e1.75) (0.59e4.81) (1.28e16.1) (0.77e5.09) (0.58e2.92)

p-value

0.59 (NS) 0.48 (NS) 0.38 (NS) 0.14 (NS) 0.13 (NS) 0.27 (NS) 0.20 (NS) 0.63 (NS) 0.15 (NS) 0.51 (NS) 0.16 (NS) 0.014 0.45 (NS) 0.40 (NS) 0.99 (NS) 0.45 (NS) 0.46 (NS) 0.13 (NS) 0.16 (NS) 0.14 (NS) 0.41 (NS) 0.17 (NS) 0.77 (NS) 0.33 (NS) 0.34 (NS) 0.12 (NS) 0.72 (NS) 0.55 (NS) 0.46 (NS) 0.84 (NS)

(0.02e9.95) (0.20e15.6) (0.47e4.33) (0.38e8.13) (0.67e23.4) (0.30e7.09) (0.16e3.95) (0.20e5.13) (1.33e29.9) (0.05e58.4) (0.57e17.4) (0.3e7.98) (0.37e24.6) (0.39e40.0) (0.57e13.0) (0.66e92.0) (0.13e14.3) (0.70e5.63) (0.26e6.85) (0.23e38.7) (0.29e13.6) (0.27e12.2) (0.41e3.26) (0.14e128.3) (0.06e6.35) (0.21e1.53) (0.59e4.81) (1.28e23.0) (0.68e5.98) (0.48e8.16)

p-value

(0.01e4.69) (0.24e16.7) (0.53e4.78) (0.17e5.39) (0.09e13.1) (0.12e7.28) (0.08e4.30) (0.14e10.1) (0.61e18.8) (0.18e57.8) (0.58e14.8) (0.10e1.46) (0.27e9.28) (0.24e48.0) (0.17e4.65) (0.22e68.4) (0.04e7.54) (0.24e8.30) (0.10e5.04) (0.16e9.02) (0.27e4.81) (0.11e14.2) (0.20e2.13) (0.15e184.4) (0.18e5.10) (0.07e1.75) (0.26e7.61) (0.67e16.0) (0.27e4.56) (0.10e5.18)

0.19 (NS) <0.0001 0.29 (NS) <0.0001 <0.0001 <0.0001 <0.0001 0.80 (NS) 0.005 0.74 (NS) 0.002 0.079 (NS) <0.0001 0.32 (NS) 0.011 0.23 (NS) <0.0001 <0.0001 0.003 <0.0001 0.003 <0.0001 0.001 0.035 0.96 (NS) <0.0001 0.22 (NS) 0.002 0.001 <0.0001

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

HLA-DRA LAG3 PVR PVRIG TIGIT CD96 CD226 TNFSF9 TNFRSF9 TNFSF18 TNFRSF18 ICOSLG ICOS OX40L OX40 CD70 CD27 TIM-3 LGALS9 PD-L1 PD-L2 PD-1 ENTPD1 IDO1 NT5E TNFRSF14 CD276 CD80 CD86 CD28

Protein

a Median (range) of gene mRNA levels; the mRNA values of the samples were normalised such that the median of the 27 head and neck normal tissues mRNA values was equal to 1. b Kruskal-Wallis H Test.

215

216

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

compared with normal head and neck tissue (p < 0.05), all being overexpressed except ICOSLG and TNFRSF14 (Table 2). Seven genes were overexpressed in more than 50% of tumours (4-1BB, GITR, ICOS, OX40L, CD70, IDO1 and CD80). OX40L had the highest mRNA level with a median of 6.52-fold and was overexpressed in 74% of tumours. Expression of PVRIG was significantly higher in HPV-positive tumours, whereas GITR, NT5E and CD276 expressions were significantly lower in HPVpositive tumours (Table 3). All immune-related genes except ICOSLG showed no modification in the expression profile in patients with HNSCC who harboured oncogene mutations (NRAS, HRAS, KRAS or PIK3CA) as compared with those with no mutation. Regarding tumour genomic alteration burden, 20 of the 30 immune-related genes (67%) were significantly overexpressed in tumours with less than three tumour genomic alterations except for GITR, which showed a lower mRNA level (Table 3). Unsupervised hierarchical clustering analyses of 96 HNSCC samples with the 30 immune-related genes showed that the majority of genes coding a receptor protein clustered together as compared with genes coding a ligand protein (Supplementary Fig. 1). 3.2.2. mRNA expression of the immune-cell subpopulation genes Among the 16 immune-cell subpopulation genes analysed, 7 genes (44%) were significantly deregulated in HNSCC tumours as compared with the normal head and neck tissue (p < 0.05). FOXP3 was overexpressed,

while only NCAM1 was underexpressed in more than 50% of the tumours as compared with the normal head and neck tissue (Table 4). Expressions of PDGFRB and NCAM1 were significantly different according to HPV status (Table 5). All immune-cell subpopulation genes except PDGFRB and FUT4 were significantly underexpressed in HNSCC with more than three tumour genomic alterations. A similar unsupervised hierarchical clustering analysis of 96 HNSCC samples with the 16 immune-cell subpopulation genes showed that the lymphocyte-specific genes also clustered together (Supplementary Fig. 2). 3.3. OX40L protein expression by immunochemistry By IHC, the OX40L protein was expressed in epithelial cancer cells for most tumour samples (H Score: 1e2.5) with a predominantly cytoplasmic location in various populations of the TME, including MICs, fibroblasts and muscle (Supplementary Fig. 3). 3.4. Prognostic value of immune gene expression Twelve of the 30 immune-related genes (40%) were associated with a short DFI in univariate analysis, including four genes with a high mRNA level and eight genes with a low mRNA level (Supplementary Table 3). High OX40L mRNA level (p Z 0.0009) and low PD-1 mRNA levels (p Z 0.004) were associated with the highest risk of recurrence (Fig. 2).

Table 4 mRNA expression of 16 immune-cell subpopulation genes relative to normal tissue mRNA level. Gene

ITGAX PDGFRB PTPRC MS4A1 CTLA4 PRF1 CD3E CD2 FOXP3 CD8A CD4 GZMA GZMB NCAM1 CD14 FUT4

Cellular specificity

Dendritic cells Fibroblast Haematopoietic cells LB LT LT LT LT LTc LTc LT helper LT/NK LT/NK NK Macrophages/monocytes Neutrophils

Alias

CD45 CD20 CD152

CD56 CD15

p-valueb

% under expression

% normal expression

% over expression

2.04 (0.52e13.0) 1.1 (0.17e5.58) 1.24 (0e6.1)

<0.0001 0.87 (NS) 0.63 (NS)

0% 3% 5%

74% 93% 83%

26% 4% 12%

1.57 (0.02e51.2) 3 (0.21e15.1) 1.35 (0.12e8.62) 1.58 (0.13e11.2) 1.62 (0.14e13.2) 3.98 (0.33e18.2) 1.46 (0.08e17.4) 1.04 (0.11e5.33) 2.16 (0.11e18.9) 2.16 (0.09e22.2) 0.15 (0.01e7.56) 1.17 (0.17e3.41) 0.7 (0.13e3.37)

0.14 (NS) <0.0001 0.13 (NS) 0.18 (NS) 0.20 (NS) <0.0001 0.65 (NS) 0.62 (NS) 0.011 0.0002 <0.0001 0.19 (NS) 0.002

22% 1% 8% 6% 8% 0% 10% 6% 8% 2% 66% 3% 12%

39% 49% 72% 68% 70% 39% 62% 90% 55% 57% 27% 95% 88%

40% 50% 20% 26% 22% 62% 28% 4% 37% 41% 7% 2% 1%

Head and neck normal tissuea

Head and neck squamous cell carcinoma

n Z 27

n Z 96

1.0 (0.22e3.23) 1.0 (0.22e5.89) 1.0 (0.39e6.41) 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

(0.06e161.9) (0.03e5.05) (0.43e3.34) (0.26e7.59) (0.19e6.73) (0e5.16) (0.21e6.54) (0.56e2.71) (0.35e6.15) (0.2e18.4) (0.1e7.6) (0.39e3.54) (0.35e4.18)

LT: T cell; LB: B cell; NK: natural killer, LTc: cytotoxic T lymphocyte. a Median (range) of gene mRNA levels; the mRNA values of the samples were normalised such that the median of the 27 head and neck normal tissues mRNA values was equal to 1. b Kruskal-Wallis H Test.

Table 5 mRNA expression of immune-cell subpopulation genes in HNSCC according to HPV status and the number of molecular alterations. Gene

Cellular specificity

Alias

HPV- verus HPVþ (84 versus 12)

ITGAX PDGFRB PTPRC MS4A1 CTLA4 PRF1 CD3E CD2 FOXP3 CD8A CD4 GZMA GZMB NCAM1 CD14 FUT4

Dendritic cells Fibroblast Haematopoietic cells LB LT LT LT LT LTc LTc LT helper LT/NK LT/NK NK Macrophages/monocytes Neutrophils

2.14 1.19 CD45 1.15 CD20 1.50 CD152 2.85 1.35 1.35 1.44 3.74 1.38 1.04 2.16 2.16 CD56 0.19 1.18 CD15 0.71

(0.52e13.0) (0.21e5.58) (0.00e6.10) (0.02e51.2) (0.21e15.1) (0.12e8.62) (0.13e11.2) (0.14e13.2) (0.33e18.2) (0.08e17.4) (0.11e5.33) (0.11e19.0) (0.09e22.2) (0.01e7.56) (0.17e3.41) (0.13e3.37)

Total number of alterations <3 versus 3 (56 vs 37)

HPVþ

p-valueb

Not mutated

Mutated

p-value

<3

1.77 0.64 1.49 2.81 3.83 1.31 2.33 2.21 5.42 1.98 1.25 2.16 2.22 0.05 1.10 0.57

0.22 (NS) 0.007 0.53 (NS) 0.80 (NS) 0.48 (NS) 0.98 (NS) 0.092 (NS) 0.11 (NS) 0.26 (NS) 0.21 (NS) 0.30 (NS) 0.60 (NS) 0.65 (NS) 0.033 0.23 (NS) 0.34 (NS)

1.98 1.13 1.14 1.43 3.00 1.32 1.35 1.44 3.97 1.33 1.04 2.11 2.16 0.16 1.12 0.70

2.20 1.07 1.44 2.05 3.20 1.57 2.07 1.87 4.53 1.98 1.05 2.16 2.73 0.12 1.29 0.85

0.33 (NS) 0.62 (NS) 0.21 (NS) 0.99 (NS) 0.68 (NS) 0.14 (NS) 0.24 (NS) 0.26 (NS) 0.63 (NS) 0.062 (NS) 0.50 (NS) 0.44 (NS) 0.52 (NS) 0.75 (NS) 0.34 (NS) 0.16 (NS)

2.35 1.19 1.58 3.18 3.86 1.57 2.30 2.34 5.13 2.34 1.23 2.55 2.38 0.19 1.33 0.76

(0.55e6.03) (0.17e1.73) (0.59e2.91) (0.10e9.06) (1.79e5.15) (0.58e4.87) (0.62e6.84) (0.61e6.78) (1.83e8.76) (0.29e10.5) (0.50e2.46) (0.50e7.85) (0.57e8.70) (0.03e1.13) (0.46e1.70) (0.29e2.66)

(0.52e13.0) (0.21e3.55) (0.00e6.10) (0.02e51.2) (0.21e15.1) (0.12e7.67) (0.13e11.2) (0.14e13.2) (0.33e18.2) (0.08e11.8) (0.11e5.33) (0.11e18.9) (0.09e22.2) (0.01e7.56) (0.17e3.41) (0.13e3.37)

(0.55e6.64) (0.17e5.58) (0.50e3.94) (0.07e8.55) (0.48e9.05) (0.40e8.62) (0.29e7.10) (0.27e8.55) (1.08e10.5) (0.38e17.4) (0.48e4.16) (0.32e16.7) (0.38e14.0) (0.03e3.12) (0.46e2.65) (0.37e2.66)

LT: T cell; LB: B cell; NK: natural killer, LTc: cytotoxic T lymphocyte. a Median (range) of gene mRNA levels; the mRNA values of the samples were normalised such that the median of the 27 head and neck normal tissues mRNA values was equal to 1. b Kruskal-Wallis H Test.

3 (0.55e13.0) (0.17e5.58) (0.00e6.10) (0.04e42.1) (0.48e15.1) (0.24e8.46) (0.29e11.2) (0.27e13.2) (1.08e18.2) (0.16e11.8) (0.37e5.33) (0.29e18.9) (0.09e22.2) (0.01e7.56) (0.40e3.41) (0.24e2.66)

1.61 0.99 0.74 0.48 1.97 1.01 0.90 0.86 2.73 0.72 0.66 1.40 1.65 0.12 0.86 0.64

p-value (0.52e6.25) (0.21e3.31) (0.11e4.96) (0.02e51.2) (0.21e7.91) (0.12e8.62) (0.13e7.10) (0.14e8.55) (0.33e10.5) (0.08e17.4) (0.11e4.16) (0.11e16.7) (0.24e14.0) (0.01e3.01) (0.17e2.06) (0.13e3.37)

0.002 0.41 (NS) <0.0001 0.0002 0.0002 0.002 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.003 0.016 0.048 0.002 0.098 (NS)

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

HPV-a

Not mutated versus mutated oncogenes (78 versus 18)

217

218

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

Disease-free interval (%)

A

Low OX40L mRNA level

High OX40L mRNA level

p=0.0009

Time (years)

Disease-free interval (%)

B

High PD-1 mRNA level

Low PD-1 mRNA level p=0.004

Time (years) Fig. 2. Relationship between disease-free interval and (A) OX40L and (B) PD-1 mRNA level.

Eight of the 16 immune-cell subpopulation genes (50%) were associated with a short DFI, including two genes with a high mRNA level and six genes with a low mRNA level (Supplementary Table 4). High PDGFRB mRNA level (p < 0.0001) and low CD3E or CD8A mRNA levels (p Z 0.0009 and p Z 0.004, respectively) were associated with the highest risk of recurrence (Fig. 3). OX40L, PD-1, PDGFRB, CD3E and CD8A immune genes were significantly associated with a short DFI (OX40L: p Z 0.005, PD-1: p Z 0.019, PDGFRB: p Z 0.0004, CD3E: p Z 0.011, and CD8A: p Z 0.016) (Supplementary Table 5) in a multivariate analysis, taking into account all clinical parameters

associated with a short DFI with a p-value <0.1 (Table 1). 3.5. Expression of actionable immune genes PD-1/PD-L1, TIGIT, OX40/OX40L and CTLA4 are currently major actionable genes in the context of immunotherapy strategies. Among the 96 tumours analysed, 82 tumours (85%) overexpressed at least one of these six genes. Thirty-three tumours (34%) simultaneously overexpressed TIGIT, CTLA4, PD-1/PD-L1 and OX40/OX40L. No tumour exclusively overexpressed PD-1 or OX40. PD-L1 was exclusively overexpressed in only one sample (1%). TIGIT was

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

Disease-free interval (%)

A

Low PDGFRB mRNA level

p<0.0001

High PDGFRB mRNA level

Time (years)

Disease-free interval (%)

B

High CD3E mRNA level

Low CD3E mRNA level p=0.0009

Time (years)

Disease-free interval (%)

C

High CD8A mRNA level

Low CD8A mRNA level p=0.004

Time (years) Fig. 3. Relationship between disease-free interval and (A) PDGFRB, (B) CD3E and (C) CD8A mRNA level.

219

220

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

Fig. 4. Clinical characteristics and molecular alterations of the main targeted actionable genes in immunotherapy. HPV, human papillomavirus.

exclusively overexpressed in only two samples (2%) and CLTA4 in three samples (3%), whereas OX40L was exclusively overexpressed in 29 samples (30%) (Fig. 4A). Fig. 4B describes the clinical and molecular characteristics of the 82 patients with HNSCC with at least one overexpressed immune gene among the six actionable genes of interest.

4. Discussion We assessed the prognostic value of selected immune gene expression in a retrospective analysis of 96 patients with HNSCC who underwent primary surgery at Institut Curie. Our results show that most significantly overexpressed genes were 4-1BB (77%), IDO1 (75%),

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

OX40L (74%) and GITR (74%) immune-related genes and FOXP3 (62%) immune-cell subpopulation genes. Eighty-five percent of tumours analysed overexpressed actionable immunity genes, including PD-1/PDL1, TIGIT, OX40/OX40L and/or CTLA4. High OX40L mRNA level (p Z 0.0009) and low PD-1 mRNA level (p Z 0.004) were associated with the highest risk of recurrence. Patients with high PDGFRB mRNA levels and low CD3E or CD8A mRNA levels also were at the highest risk of recurrence. Overall, around half of immune genes had a deregulated mRNA level in tumour cells as compared with the normal head and neck tissue. Tumours with a low number of genomic alterations had higher mRNA levels of immune genes. More specifically, our results show that among immune-cell subpopulation genes, antigen-presenting cells, dendritic cells and B and T cells had a higher mRNA level in tumours, whereas natural killer cellespecific genes (i.e. NCAM1) were underexpressed. Furthermore, we observed a high FOXP3 mRNA level not correlated to recurrence. This suggests the high number of T regulatory cells that was also reported in patients with HNSCC [10]. Positive associations were observed between OX40L and FOXP3 mRNA levels and between OX40L and CD4 mRNA levels using the Spearman rank correlation test (r Z þ0.27, p Z 0.0083 and r Z þ0.46, p < 0.0001, respectively). OX40L, PD-1, PDGFRB, CD3E and CD8A were associated with a poor prognosis in our study, especially a high mRNA level of OX40L. The prognostic value of OX40L expression is controversial in the literature and depends on the type of cancer [19,20]. The overexpression of OX40L was significantly associated with a higher risk of recurrence in bladder cancer [19], whereas it was associated with prolonged progressionfree survival in glioblastoma [20]. In our patients with HNSCC, OX40L was expressed in the microenvironment of the tumour cell notably in fibroblasts at the protein level. In this regard, we observed a marked positive association between OX40L and PDGFRB mRNA levels using the Spearman rank correlation test (r Z þ0.47 and p < 0.0001, data not shown). Similarly, the expression of OX40L was detected in a subset of carcinoma-associated fibroblasts characterised by an immunosuppressive environment in patients with triple-negative breast cancer [21]. No clear cell location of the OX40L protein was reported in the literature for HNSCC. Low mRNA level of PD-1 correlated with a poor prognosis in our series. The absence of PD-L1 mRNA overexpression in circulating tumour cells after antiePD1 treatment was strongly associated with an objective response in patients with HNSCC [22]. The prognostic value of PD-1 mRNA level was also reported in highgrade serous ovarian carcinoma [23] and in non-small cell lung cancer [24]. Low mRNA level of PD-L1 also had a poor prognosis although less significant than PD-1 which

221

is consistent to KEYNOTE-048 trial (NCT02358031) interim findings that reported an improved overall survival and duration of response versus standard therapy in patients with PD-L1epositive recurrent or metastatic HNSCC [25]. Our results also demonstrated that high mRNA level of PDGFRB, a major fibroblast-specific gene, was associated with a poor prognosis. Similar results were reported in prostate cancer [26]. A high PDGFRB protein expression was also reported to correlate with a short survival in patients with renal cell carcinoma [27] and pancreatic adenocarcinoma [28]. Patients with HNSCC with low mRNA levels of CD3E or CD8A had a poor prognosis. Similarly, high CD3 and CD8 mRNA expression was associated with a decreased risk of relapse in patients with early breast cancer [29]. CD8 overexpression correlated with prolonged overall survival and recurrence-free survival in patients with bladder cancer [19]. Eighty-five percent of HNSCC tumours had at least an actionable immune gene overexpression (PD-1/PDL1, TIGIT, OX40L/OX40, and CTLA4) in our series. OX40L mRNA level was overexpressed in 74% of tumours and had the worst prognostic value, suggesting that OX40L is a relevant therapeutic target for patients with HNSCC. Several clinical trials evaluate OX40L agonists either as single agent or in combination with antiePD-1/PDL1 agents (NCT02315066, NCT02923349, NCT02410512, NCT02221960, NCT02705482). A phase I trial evaluating an anti-OX40 agonistic monoclonal antibody (9B12) in 30 patients with refractory metastatic solid malignancies showed a favourable safety profile and tumour shrinkage in 12 patients [30]. In addition in our series, 34% of HNSCC tumours coexpressed PD-1/PD-L1, TIGIT, OX40/OX40L and CTLA4, suggesting that a combination of immune checkpoint inhibitors may be a relevant therapeutic strategy in patients with HNSCC. Several clinical trials are ongoing with other immunotherapy drugs such as GITR (e.g. NCT01239134 or NCT02697591) or 4-1BB (e.g. NCT03364348) generally administered in combination in advanced solid malignancies.

5. Conclusions OX40L and PDGFRB overexpression was associated with poor outcome in patients with HNSCC treated with primary surgery. On the contrary, PD-1 overexpression was associated with good prognosis. These results suggest their relevance as potential prognostic biomarkers to be validated in an independent cohort. Immunotherapy was recently demonstrated to improve overall survival in the recurrent and/or metastatic setting of patients with HNSCC with pembrolizumab [9] and nivolumab [8] that target PD-1. However, only a

222

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223

minority of patients benefit from these agents [8,9], highlighting the urgent need to identify other relevant immune targets.

[8]

Funding sources [9]

This work was supported by the Fondation ARC pour la recherche sur le cancer (ARC) and ICGEx Project ANR-10-EQPX-03 (Equipement de biologie inte´grative du cancer pour une me´decine personnalise´e). [10]

Conflict of interest statement C.L.T. has participated in advisory boards of MSD, BMS, Merck Serono, Astra Zeneca, Novartis, Roche and Nanobiotix. All other authors report no conflict of interest.

Acknowledgements We would like to acknowledge the help of the NR-10IDEX-0001-02 PSL*, ANR-11-LABX-0043 and CIC IGR-Curie 1428 (Center Of Clinical Investigation) for their support. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.ejca.2019.08.028.

[11]

[12]

[13]

[14]

[15]

References [1] Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN. 2012. PubMed e NCBI n.d, https://www.ncbi.nlm.nih.gov/pubmed/25220842. [Accessed 22 December 2017]. [2] Fakhry C, Westra WH, Li S, Cmelak A, Ridge JA, Pinto H, et al. Improved survival of patients with human papillomaviruspositive head and neck squamous cell carcinoma in a prospective clinical trial. J Natl Cancer Inst 2008;100:261e9. https: //doi.org/10.1093/jnci/djn011. [3] Cancer Genome Atlas Network. Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature 2015;517:576e82. https://doi.org/10.1038/nature14129. [4] Sablin M-P, Dubot C, Klijanienko J, Vacher S, Ouafi L, Chemlali W, et al. Identification of new candidate therapeutic target genes in head and neck squamous cell carcinomas. Oncotarget 2016;7:47418e30. https: //doi.org/10.18632/oncotarget.10163. [5] Dubot C, Bernard V, Sablin MP, Vacher S, Chemlali W, Schnitzler A, et al. Comprehensive genomic profiling of head and neck squamous cell carcinoma reveals FGFR1 amplifications and tumour genomic alterations burden as prognostic biomarkers of survival. Eur J Cancer Oxf Engl 1990 2018;91:47e55. https: //doi.org/10.1016/j.ejca.2017.12.016. [6] Bonner JA, Harari PM, Giralt J, Azarnia N, Shin DM, Cohen RB, et al. Radiotherapy plus cetuximab for squamous-cell carcinoma of the head and neck. N Engl J Med 2006;354:567e78. https://doi.org/10.1056/NEJMoa053422. [7] Vermorken JB, Mesia R, Rivera F, Remenar E, Kawecki A, Rottey S, et al. Platinum-based chemotherapy plus cetuximab in

[16]

[17]

[18] [19]

[20]

[21]

[22]

head and neck cancer. N Engl J Med 2008;359:1116e27. https: //doi.org/10.1056/NEJMoa0802656. Ferris RL, Blumenschein G, Fayette J, Guigay J, Colevas AD, Licitra L, et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med 2016;375:1856e67. https: //doi.org/10.1056/NEJMoa1602252. Seiwert TY, Burtness B, Mehra R, Weiss J, Berger R, Eder JP, et al. Safety and clinical activity of pembrolizumab for treatment of recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol 2016;17:956e65. https: //doi.org/10.1016/S1470-2045(16)30066-3. Badoual C, Hans S, Rodriguez J, Peyrard S, Klein C, Agueznay NEH, et al. Prognostic value of tumor-infiltrating CD4þ T-cell subpopulations in head and neck cancers. Clin Cancer Res Off J Am Assoc Cancer Res 2006;12:465e72. https: //doi.org/10.1158/1078-0432.CCR-05-1886. Pretscher D, Distel LV, Grabenbauer GG, Wittlinger M, Buettner M, Niedobitek G. Distribution of immune cells in head and neck cancer: CD8þ T-cells and CD20þ B-cells in metastatic lymph nodes are associated with favourable outcome in patients with oro- and hypopharyngeal carcinoma. BMC Cancer 2009;9: 292. https://doi.org/10.1186/1471-2407-9-292. Hartmann E, Wollenberg B, Rothenfusser S, Wagner M, Wellisch D, Mack B, et al. Identification and functional analysis of tumor-infiltrating plasmacytoid dendritic cells in head and neck cancer. Cancer Res 2003;63:6478e87. Kerrebijn JD, Balm AJ, Knegt PP, Meeuwis CA, Drexhage HA. Macrophage and dendritic cell infiltration in head and neck squamous-cell carcinoma; an immunohistochemical study. Cancer Immunol Immunother CII 1994;38:31e7. Germain C, Gnjatic S, Dieu-Nosjean M-C. Tertiary lymphoid structure-associated B cells are key players in anti-tumor immunity. Front Immunol 2015;6:67. https: //doi.org/10.3389/fimmu.2015.00067. Montler R, Bell RB, Thalhofer C, Leidner R, Feng Z, Fox BA, et al. OX40, PD-1 and CTLA-4 are selectively expressed on tumor-infiltrating T cells in head and neck cancer. Clin Transl Immunol 2016;5:e70. https://doi.org/10.1038/cti.2016.16. Forbes SA, Bhamra G, Bamford S, Dawson E, Kok C, Clements J, et al. The catalogue of somatic mutations in cancer (COSMIC). Curr Protoc Hum Genet 2008. https: //doi.org/10.1002/0471142905.hg1011s57 [Chapter 10]:Unit 10.11. Pignot G, Bieche I, Vacher S, Gu¨et C, Vieillefond A, Debre´ B, et al. Large-scale real-time reverse transcription-PCR approach of angiogenic pathways in human transitional cell carcinoma of the bladder: identification of VEGFA as a major independent prognostic marker. Eur Urol 2009;56:678e88. https: //doi.org/10.1016/j.eururo.2008.05.027. Cox DR. Regression models and life-tables. J R Stat Soc Ser B Methodol 1972;34:187e220. Le Goux C, Vacher S, Pignot G, Sibony M, Barry Delongchamps N, Terris B, et al. mRNA Expression levels of genes involved in antitumor immunity: identification of a 3-gene signature associated with prognosis of muscle-invasive bladder cancer. OncoImmunology 2017;6:e1358330. https: //doi.org/10.1080/2162402X.2017.1358330. Shibahara I, Saito R, Zhang R, Chonan M, Shoji T, Kanamori M, et al. OX40 ligand expressed in glioblastoma modulates adaptive immunity depending on the microenvironment: a clue for successful immunotherapy. Mol Cancer 2015;14: 41. https://doi.org/10.1186/s12943-015-0307-3. Costa A, Kieffer Y, Scholer-Dahirel A, Pelon F, Bourachot B, Cardon M, et al. Fibroblast heterogeneity and immunosuppressive environment in human breast cancer. Cancer Cell 2018;33: 463e79. https://doi.org/10.1016/j.ccell.2018.01.011. e10. Strati A, Koutsodontis G, Papaxoinis G, Angelidis I, Zavridou M, Economopoulou P, et al. Prognostic significance of PD-L1

C. Lecerf et al. / European Journal of Cancer 121 (2019) 210e223 expression on circulating tumor cells in patients with head and neck squamous cell carcinoma. Ann Oncol Off J Eur Soc Med Oncol 2017;28:1923e33. https://doi.org/10.1093/annonc/mdx206. [23] Darb-Esfahani S, Kunze CA, Kulbe H, Sehouli J, Wienert S, Lindner J, et al. Prognostic impact of programmed cell death-1 (PD-1) and PD-ligand 1 (PD-L1) expression in cancer cells and tumor-infiltrating lymphocytes in ovarian high grade serous carcinoma. Oncotarget 2016;7:1486e99. https: //doi.org/10.18632/oncotarget.6429. ´ , Estors M, Martı´nez[24] Lafuente-Sanchis A, Zu´n˜iga A Herna´ndez NJ, Cremades A, Cuenca M, et al. Association of PD-1, PD-L1, and CTLA-4 gene expression and clinicopathologic characteristics in patients with non-small-cell lung cancer. Clin Lung Cancer 2017;18:e109e16. https: //doi.org/10.1016/j.cllc.2016.09.010. [25] Burtness B, Harrington KJ, Greil R, Soulie`res D, Tahara M, De Castro G, et al. LBA8_PRKEYNOTE-048: phase III study of first-line pembrolizumab (P) for recurrent/metastatic head and neck squamous cell carcinoma (R/M HNSCC). Ann Oncol 2018; 29. https://doi.org/10.1093/annonc/mdy424.045.

223

[26] Singh D, Febbo PG, Ross K, Jackson DG, Manola J, Ladd C, et al. Gene expression correlates of clinical prostate cancer behavior. Cancer Cell 2002;1:203e9. [27] Fro¨din M, Mezheyeuski A, Corvigno S, Harmenberg U, Sandstro¨m P, Egevad L, et al. Perivascular PDGFR-b is an independent marker for prognosis in renal cell carcinoma. Br J Cancer 2017;116:195e201. https://doi.org/10.1038/bjc.2016.407. [28] Yuzawa S, Kano MR, Einama T, Nishihara H. PDGFRb expression in tumor stroma of pancreatic adenocarcinoma as a reliable prognostic marker. Med Oncol Northwood Lond Engl 2012;29:2824e30. https://doi.org/10.1007/s12032-012-0193-0. [29] Tsiatas M, Kalogeras K, Manousou K, Wirtz R, Gogas H, Veltrup E, et al. Abstract P1-07-03: evaluation of the prognostic value of CD3, CD8 and FOXP3 mRNA expression in early breast cancer patients treated with anthracycline-based adjuvant chemotherapyvol. 78; 2018. https://doi.org/10.1158/1538-7445.SABCS17-P1-07-03. [30] Curti BD, Kovacsovics-Bankowski M, Morris N, Walker E, Chisholm L, Floyd K, et al. OX40 is a potent immune-stimulating target in late-stage cancer patients. Cancer Res 2013;73:7189e98. https://doi.org/10.1158/0008-5472.CAN-12-4174.